ComfyUI Text-to-Video AI Converter: Create Videos Like Runway Gen-3 with 8GB VRAM

ComfyUI-CogVideoX-MZ: Text-to-Video AI with 4-bit Quantization (Update 2024-09-05)

Key Features:

Memory Efficiency: Uses less than 8GB of VRAM, making it accessible for users with limited hardware resources.Visit github.com
4-bit Quantization: Optimized for performance with select layers quantized to 4 bits, enhancing speed while maintaining quality.
DPM Sampling Mode: Uses Denoising Diffusion Probabilistic Model (DPM) for improved sampling precision.
GGUF Weight Dtype: Supports the efficient GGUF format for weight_dtype to handle large-scale models.
Resolution: Outputs videos at 480×720 with 32 frames, suitable for high-quality video generation with low resource consumption.

Start ComfyUI on your system.
Once ComfyUI is open, navigate to the ComfyUI Manager. This is typically located in the interface under settings or tools.
Inside the ComfyUI Manager, find the Custom Node Manager section. This is where you’ll manage additional nodes for ComfyUI.
In the search bar of the Custom Node Manager, type
```
CogVideoXWrapper
```
to locate the wrapper for the CogVideoX model.
Once you find CogVideoXWrapper in the search results, click on the install button.
After installation is complete, restart ComfyUI to apply the changes and load the new node

How to Set Up the CogVideoX-5B Model:

Download the Model:
- Download the model file
```
CogVideoX_5b_GGUF_Q4_0.safetensors
```
  and place it in the
```
unet
```
  folder.
- Additionally, download the workflow image from
```
examples/workflow_GGUF_Q4_0.avif
```
  for easy setup.
Required Dependencies:
- Install the model loader from CogVideoXWrapper for seamless integration with ComfyUI.
Additional Downloads:
- Download
```
CogVideoX_5b_fp8_4em.safetensors
```
  to the
```
unet
```
  folder.
- Download
```
diffusion_pytorch_model.safetensors
```
  to the
```
vae
```
  folder (rename it for better organization).
- Download
```
t5xxl_fp8_e4m3fn.safetensors
```
  to the
```
clip
```
  folder (if you already have it from SD3/Flux, no need to download again).

Final Workflow Setup:

The complete setup process, including the workflow diagram, can be found in
```
examples/workflow.avif
```
.

This guide makes it easy to configure CogVideoX-5B for creating high-quality text-to-video content with ComfyUI, even with 8GB of VRAM.